Browse Prior Art Database

Cloud Workload Routing by Temperature Vectorization Disclosure Number: IPCOM000244530D
Publication Date: 2015-Dec-18
Document File: 7 page(s) / 203K

Publishing Venue

The Prior Art Database


Disclosed is a data center cooling technology that works through workload management in a cloud environment and uses heatmaps to handle temperature variations or failure scenarios.

This text was extracted from a PDF file.
This is the abbreviated version, containing approximately 36% of the total text.

Page 01 of 7

Cloud Workload Routing by Temperature Vectorization

Within the cloud computing industry, one concern is the overall temperature in data centers. Information Technology (IT) organizations and centers continually search for ways to improve the use of the cooling energy to achieve better performance while increasing processing power and maintaining the correct lifespan of the servers. Hence, more stable cooling systems can imply the maximization of the use of computing resources.

To address the temperature issue, current hot/cold aisle designs are not only the most adopted by the industry, but also the most excelled in this field. Modern IT equipment has adopted these strategies to achieve desired temperatures in data centers . These approaches isolate airflow (hot or cold) in order to reduce power consumption while increasing the airflow within the data center through density of cold/hot air.

However, this current cooling technology only focuses on the overall temperature of data centers. In a cloud environment, depending on the workload, a specific machine can gradually reach higher temperatures than others can, without significantly changing the overall temperature of the data center instantaneously. Moreover, in failure scenarios, available solutions focus on the maintenance of cooling air as long as possible, up to the point at which temperatures reach intolerable levels, and hence, the entire systems shut down. These solutions do not consider the workload handled by these machines. The primary focus is the heat of the hardware structure, as opposed to the origin of the heat expelled by the machines.

Figure 1: Data center

While observing Figure 1, consider the following scenario: suppose that rack 1 increased its temperature more than the other racks due to workload unbalancing. Somehow, rack 1 was assigned to process much more work than the other racks.


Page 02 of 7

Even though cold airflow is properly reaching rack 1, the tendency of the controllers embedded in the High Ventilation Air Conditioning (HVAC) units is to actuate even more in order to stabilize the temperature variations, according to simple control theory. Therefore, with the cost of more electric energy, alterations in cold airflow are produced at the region of rack 1, because it is the main cause of the overall temperature increase.

In the event of a complete failure scenario, all racks increase in temperature due to an HVAC failure, and hence, the mechanism acts in the same manner as mentioned before. However, this is a more circumstantial situation, in which the machines probably start shutdown procedures.

The solution is a data center cooling technology that works through workload management in a cloud environment and uses heatmaps to handle temperature variations or failure scenarios. The core novel aspects of the solution follow:

A new fine-grained variable to route new workloads and virtual machine allocation based on the data center heat map Gr...